Unmanned aerial vehicle inspection adaptive planning method for highway construction supervision

Abstract

The application discloses a method for adaptive planning of unmanned aerial vehicle inspection for highway construction supervision, and relates to the technical field of traffic control, comprising the following steps: proposing the unmanned aerial vehicle inspection and supervision content for highway construction, constructing dynamic division of inspection sub-regions based on supervision requirements of each construction stage, adaptive optimization of unmanned aerial vehicle hovering time based on section hidden danger event supervision requirements, dynamic linkage scheduling of unmanned aerial vehicles based on occurrence and end of emergent events; the application can dynamically divide the highway construction supervision inspection sub-regions according to the supervision content and the maximum endurance of the unmanned aerial vehicle; the hovering time is dynamically optimized in combination with the influence of various hidden danger events according to the supervision content of different construction stages of the highway; the application also has emergent event emergency response capability; through adaptive planning of dynamic adjustment of the unmanned aerial vehicle inspection range, the hovering time and emergency scheduling, the adaptive adjustment of the unmanned aerial vehicle in the highway construction supervision is realized.

Description

Technical Field

[0001] This invention relates to the field of traffic control technology, and in particular to an adaptive planning method for unmanned aerial vehicle (UAV) inspections for highway construction supervision. Background Technology

[0002] As the scale of highway construction continues to expand and the construction environment becomes increasingly complex, it places extremely high demands on progress control and safety supervision during construction. Currently, highway construction supervision mainly relies on manual inspections and fixed video surveillance. The former is inefficient, costly, and difficult to cover dangerous areas such as high slopes, while the latter cannot achieve comprehensive supervision due to its large blind spots and poor deployment flexibility.

[0003] In recent years, with the booming development of the low-altitude economy, drones have been gradually introduced into the field of highway construction supervision due to their advantages of maneuverability, wide field of view, and ability to carry multiple sensors. Currently, drones in highway construction supervision mainly adopt two modes: manual flight or automatic flight along preset fixed routes. Among them, manual flight allows for flexible adjustment of the viewing angle according to the on-site situation, while fixed route flight realizes the automation and repeatability of the inspection path, which improves the efficiency of supervision to a certain extent. However, the existing application methods still have obvious limitations: Limitation 1: Fixed inspection range. The inspection range is usually set manually and cannot be adaptively adjusted according to the supervision needs of different construction stages. Limitation 2: No hovering time or fixed hovering time. Most drones fly directly over for patrol, making it impossible to conduct long-term or timed observations based on the monitoring needs of different locations. A few drones will hover at fixed points, but the hovering time is usually set to a fixed value and cannot be optimized according to the dynamic changes of potential hazards in different sections of the construction site. Limitation 3: Lack of emergency dispatch capability for sudden events. When a sudden event occurs, it is impossible to dispatch drones to the scene for inspection in a timely manner.

[0004] These shortcomings make it difficult for drones to adapt to the dynamic evolution of construction procedures, the dynamic changes of potential hazards, and the emergency response needs of sudden events. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide an adaptive planning method for drone inspections in highway construction supervision, which can dynamically adjust the drone inspection range, hovering time, and emergency dispatch according to the real-time changes in highway construction supervision needs.

[0006] To address the problems of the prior art, the technical solution of the present invention is as follows: An adaptive planning method for drone inspection in highway construction supervision includes the following steps: S1. Propose drone inspection and supervision content for highway construction. Taking the construction scenario of new highway subgrade and pavement as the application object, five construction stages are selected as the supervision objects: highway clearing and demolition, subgrade construction, pavement base construction, pavement surface construction, and traffic safety electromechanical construction. When the drone performs inspection tasks, it monitors the key equipment and personnel operation behavior in each construction stage in real time. According to the severity of the hidden dangers, they are divided into three categories: major safety hazards, general safety hazards, and non-standard operations. The sections along the highway construction route are divided into sections according to the coverage range of a single drone hovering as the basic detection unit. The key equipment includes at least one of the following: excavator, dump truck, grader, road roller, paver, joint cutter, pile driver, and core drill.

[0007] S2. Construct a dynamic division of inspection sub-zones based on the supervision needs of each construction stage. With the maximum flight time of the drone as a constraint, and starting from the location of the drone nest, the hovering time of each segment, the flight time between adjacent segments, the flight time from the drone nest to the first segment, and the flight time from the last segment back to the drone nest are accumulated segment by segment along the construction route. When the total accumulated time first approaches but does not exceed the maximum flight time of the drone, the boundary of the current sub-region is defined. The remaining undefined segments are divided successively until the entire route is covered. Each sub-region is equipped with a main drone to perform routine inspection tasks and emergency inspection tasks, and a backup drone is equipped to quickly take over when the main drone's power is insufficient or in case of emergency dispatch. The specific method for dividing the inspection sub-areas is as follows: the inspection altitude of the drone is set to "40-50 meters". At this altitude, the drone covers an area with a cross-sectional width of 50 meters and a road length of 70 meters. This area is used as a segment and continuously deployed along the construction route. Starting from the drone's nest position, the maximum hovering time and flight time are accumulated segment by segment until the accumulated time first approaches but does not exceed the maximum endurance of the drone. When this time, the boundary of the current sub-area is defined. The remaining undefined segments continue to repeat the above division process until the entire road segment is divided.

[0008] For road sections less than 70 meters long that span multiple construction phases, due to the different regulatory requirements and hovering times at different construction phases, these road sections are inspected as independent sections and are not combined with other road sections.

[0009] S3. Adaptive optimization of drone hovering time based on the needs of monitoring potential hazards in specific areas. When the drone is first inspected, the hovering time of each section is set to the minimum hovering time of the corresponding construction stage. In subsequent iterations and optimizations, the hovering time is continuously optimized based on the previous hovering time. When the number of potential safety incidents increases in a certain section, the system iterates sequentially according to the priority order of major safety hazards, general safety hazards, and non-standard operations, calculating the hovering time increment and updating it each time. When the number of potential hazards in a certain section decreases, the smallest values ​​are selected from the incremental contribution values ​​of similar events in the historical records of that section and deducted. S4. Dynamic scheduling of drones based on the occurrence and termination of emergencies. S41. When an emergency occurs during highway construction, an emergency dispatch strategy shall be adopted based on the current status and remaining power of the main UAV. S42. After the emergency ends, the interruption scenarios will be classified according to the relationship between the current position of the drone and the inspection progress, and the optimal inspection path will be planned for each scenario.

[0010] Furthermore, each designated sub-area is equipped with one main inspection drone and an additional backup drone to quickly take over when the main drone's power is insufficient to complete the remaining tasks or in the event of an emergency.

[0011] Furthermore, the monitoring equipment carried by the drone includes a high-definition zoom camera, a thermal imager, and a lidar.

[0012] Furthermore, the hovering time of each section calculated in step S2 is the maximum hovering time corresponding to each section, which is selected based on the preset hovering time range of each construction stage.

[0013] Furthermore, the formula for dynamically dividing the inspection sub-area in step S2 is as follows: ; in, The flight time for the drone to reach the first segment; The number of regulatory sections within a sub-region; For the first Maximum hovering time for each segment; For the first Section 1 to 2 Flight time for each segment; The flight time for the drone to return to base; The maximum flight time of the main inspection drone.

[0014] Furthermore, during the initial inspection of the drone in step S3, the hovering time for each section is set to the minimum hovering time for the corresponding construction stage, based on the preset hovering time range for each construction stage.

[0015] Furthermore, the hovering time iteration rule when the number of potential hazards increases in step S3 is as follows: Let the first Before the next inspection, the hovering time of a certain section was: This time, a major new safety hazard has been added. Individual, general safety hazards One, non-standard operation One; iterate through each potential hazard event in order of priority, calculating the increment for each iteration, as shown in the following formula: ; in, The weight of the potential hazards currently being addressed; Maximum hovering time; The hovering time before the iteration; The current hover time is updated immediately after each iteration. After processing all newly added potential events, the final hover time is calculated using the following formula: ; in, For the first The hovering time during the next inspection; For the first The hovering time during the next inspection; They represent the first One major safety hazard, the first A general safety hazard, the first The specific incremental value generated during the iteration process by an irregular operation.

[0016] Furthermore, the hovering time deduction rule for reducing the number of potential hazards in step S3 is as follows: Let the first During the second inspection, it was confirmed that major safety hazards had been reduced. Individual, general safety hazards One, non-standard operation One; from the set of incremental values ​​of similar hidden danger events in the historical records of this section, select the smallest one respectively. The increments are summed and then subtracted from the current hover time, using the following formula: ; in, The smallest in history The sum of the incremental values; The smallest in history The sum of the incremental values; The smallest in history The sum of the incremental values.

[0017] Furthermore, step S41, emergency dispatch based on sudden events, includes three dispatch scenarios: Scenario 1: When an emergency occurs, the main drone is in the nest. Directly dispatch the main drone to the scene of the emergency. In real time, determine whether the event has ended. If the event has ended, execute the return-to-home operation. If the event has not ended, determine whether the main drone has enough power left to return to home. If the power is only enough to return to home, return to home immediately. If the power is sufficient, continue to maintain the on-site monitoring status until the event ends before returning to home.

[0018] Scenario 2: When an emergency occurs, the main drone is not in its nest and its remaining battery power is insufficient to meet the emergency's power threshold requirements. The main drone continues to perform its regular inspection tasks, while a backup drone is dispatched to take off and fly to the emergency site. The backup drone will determine in real time whether the emergency has ended. If the emergency has ended, the backup drone will return to base. If the emergency has not ended, the backup drone will determine whether its remaining battery power is sufficient for a return. If the battery power is insufficient, the backup drone will return directly. If the battery power is sufficient, the backup drone will continue on-site monitoring until the emergency ends and then return to base.

[0019] Scenario 3: When an emergency occurs, if the main drone is not in its nest and its remaining battery power is sufficient to meet the emergency's power threshold requirements, the routine inspection mission is interrupted, and the drone prioritizes flying to the emergency site for testing. If the emergency is not over, determine if the main drone's remaining battery power is only enough for a return trip. If the battery power is insufficient, the main drone returns directly, and a backup drone is dispatched to take over the remaining routine inspection mission. If the battery power is sufficient, on-site testing continues. If the emergency has ended, determine if the main drone's remaining battery power is sufficient to meet the requirements for inspecting the remaining section and returning to the nest. If the battery power is insufficient, the main drone returns, and a backup drone is dispatched to take over the remaining inspection mission. If the battery power is sufficient, the main drone continues to complete the remaining routine inspection mission.

[0020] Furthermore, the inspection path planning after the emergency ends in step S42 is divided into three optimization strategies based on the drone interruption scenario: Current section interruption scenario: Dispatch the drone to continue moving forward from the current section and complete the inspection tasks of the remaining uninspected sections in sequence; Scenario of interruption in inspected section: Dispatch the drone forward and complete the inspection tasks of the uninspected section in sequence at the interruption point; Scenario of interruption in uninspected section: Calculate the total time for both "forward" and "backward" paths, and select the optimal path with the shortest time to complete the remaining inspection tasks; The total time for the "forward" path is the sum of the flight time from the emergency segment to the final segment, the flight time from the final segment to the interrupted segment, and the flight time from the interrupted segment to the nest; the total time for the "backward" path is the sum of the flight time from the emergency segment to the interrupted segment, the flight time from the interrupted segment to the final segment, and the flight time from the final segment to the nest.

[0021] Compared with the prior art, the advantages of the present invention are as follows: 1. This invention targets routine inspection tasks. Based on the regulatory content of different construction stages and combined with the maximum endurance of drones, it dynamically divides highway construction supervision and inspection sub-zones. As the construction progresses, the size of the inspection sub-zones is dynamically adjusted to achieve precise matching between the inspection range and the construction progress.

[0022] 2. This invention can monitor different construction stages of highways and dynamically iterate the hovering time based on the hovering time of each section in the sub-zone during the previous inspection, as well as the severity and number of potential incidents. This continuous optimization of hovering time achieves precise dynamic matching between hovering time and the risk level of highway construction sites.

[0023] 3. This invention possesses emergency response capabilities for sudden events. When a sudden event occurs in the construction area, emergency inspection tasks are carried out, and the dynamic linkage scheduling algorithm of UAVs is used to realize the real-time detection of the sudden event. After the sudden event ends, the inspection path of the UAVs for the remaining sections is planned based on the remaining battery life of the UAVs and the detection needs of the remaining sections. This ensures that the sudden event is responded to in a timely manner, while minimizing the interruption time of regular inspections, and realizing the coordinated and efficient operation of emergency inspections and regular inspections. Attached Figure Description

[0024] Figure 1 This is the overall flowchart of the UAV inspection adaptive planning method for highway construction supervision according to the present invention.

[0025] Figure 2 These are schematic diagrams illustrating different construction stages.

[0026] Figure 3 This is a diagram of the equipment used during the site clearing and demolition phase.

[0027] Figure 4 This is a diagram of the equipment used during the roadbed construction phase.

[0028] Figure 5 This is a diagram of the equipment used during the road base construction phase.

[0029] Figure 6 This is a diagram of the equipment used in the road surface layer construction stage.

[0030] Figure 7 These are diagrams of equipment used during the construction phase of traffic safety electromechanical systems.

[0031] Figure 8 A schematic diagram of the drone inspection routes for different sub-regions.

[0032] Figure 9 This is a flowchart of the emergency dispatch process for drones in the event of a sudden incident.

[0033] Figure 10 This is a flowchart for drone inspection path planning at the end of an emergency.

[0034] Figure 11 This is a diagram showing the location of an emergency in the current inspection section.

[0035] Figure 12 This is a diagram illustrating an emergency occurring in an inspected section.

[0036] Figure 13 This is a diagram illustrating an uninspected section of a region where an emergency has occurred. Detailed Implementation

[0037] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0038] An adaptive planning method for drone inspection in highway construction supervision includes the following steps: S1. Propose drone inspection and supervision content for highway construction. Five construction stages of highways—clearing and demolition, roadbed construction, pavement base course, pavement surface course, and traffic safety and electromechanical systems—were selected as the monitoring targets. Drones equipped with monitoring devices were used to inspect potential hazards at each construction stage. The severity of potential hazards was categorized into three types: major safety hazards, general safety hazards, and non-standard operations. S2. Construct a dynamic division of inspection sub-zones based on the supervision needs of each construction stage. Constrained by the maximum flight time of the drone, and starting from the location of the drone nest, the hovering time of each segment, the flight time between adjacent segments, the flight time from the drone nest to the first segment, and the flight time from the last segment back to the drone nest are accumulated segment by segment along the construction route. When the total accumulated time first approaches but does not exceed the maximum flight time of the drone, the boundary of the current sub-region is defined. The remaining undefined segments are then divided until the entire road segment is covered. S3. Adaptive optimization of drone hovering time based on the needs of monitoring potential hazards in specific areas. Based on the hovering time of each section and the severity and number of potential hazards in the previous inspection, when the number of potential hazards increases, the hovering time increment corresponding to each potential hazard is calculated iteratively according to the priority of the potential hazards, the hovering time is increased step by step, and the increment contribution value of the event is recorded. When the number of potential incidents decreases, deduct the minimum incremental contribution values ​​of similar incidents in the historical records of that section. S4. Dynamic scheduling of drones based on the occurrence and termination of emergencies. S41. In response to emergencies, based on the drone's current remaining battery power, determine whether to continue executing the regular inspection route or interrupt the regular inspection mission, and prioritize flying to the emergency location for emergency inspection. S42. After the emergency ends, based on the remaining flight time of the drones and the inspection needs of the remaining sections, plan the inspection path for the drones in the remaining sections.

[0039] The more detailed technical solution is as follows: like Figure 1 As shown, the inspection and supervision targets are firstly the five stages of highway clearing and demolition, roadbed construction, road base layer, road surface layer and traffic safety electromechanical systems; the drone is equipped with a high-definition zoom camera, thermal imager and lidar to identify three types of hidden dangers in real time during the inspection: major safety hazards, general safety hazards and non-standard operations.

[0040] In the routine inspection planning for highway construction, the maximum flight time of the drone is used as a constraint. Combined with the monitoring time required for different construction stages of the highway, the maximum hovering time required for each section, the flight time between adjacent sections, and the round-trip time of the drone are accumulated along the construction route from the drone's nest location until the total accumulated time is close to but does not exceed the flight time limit. Based on this, the inspection sub-area is dynamically delineated.

[0041] Based on this, adaptive iterative optimization is performed on the hovering time of each section during highway construction inspection. When an increase in potential safety events is detected, the system iterates sequentially according to the priority of major safety hazards, general safety hazards, and non-standard operations. For each new potential safety event, the corresponding hovering time increment is calculated and recorded as a contribution value. When a decrease in potential safety events is detected, the smallest contribution values ​​of similar events in the historical records of that section are selected and deducted, i.e., the sum of these smallest increments is subtracted from the current hovering time.

[0042] When an emergency occurs during highway construction, an emergency inspection plan is developed. Based on the remaining battery power of the main drone, it is determined whether to continue the regular inspection mission or interrupt the regular inspection mission and prioritize flying to the location of the emergency for emergency inspection. After the emergency ends, the inspection path of the drone for the remaining section is planned based on the remaining battery life of the main drone and the inspection needs of the remaining section.

[0043] The above method achieves adaptive adjustment of drone inspection in highway construction supervision by dynamically adjusting the drone inspection range, hovering time and emergency dispatch.

[0044] 1. Propose the content of drone inspection and supervision for highway construction. like Figure 2As shown, a typical highway new roadbed and pavement construction scenario is selected as the target of drone supervision. The process is divided into five stages: clearing and demolition, roadbed construction (including six processes: filling, paving, compaction, inspection, drainage ditches, and slope protection), pavement base construction (including three processes: paving, compaction, and maintenance), pavement surface construction (including four processes: paving, compaction, joint cutting and sealing, and inspection), and traffic safety and electromechanical construction (guardrails, signs, and markings). Drones are used for inspection and supervision.

[0045] like Figures 3 to 7 As shown in Tables 1 to 5, the equipment used by drones at different construction stages and the content requiring supervision (only common supervision content is listed here) are categorized into three types based on the severity of potential safety hazards: major safety hazards, general safety hazards, and non-standard operations. When performing inspection tasks, drones use high-definition zoom cameras, thermal imagers, and lidar to monitor in real time the key equipment (excavators, dump trucks, water trucks, graders, vibratory rollers, tire rollers, pavers, asphalt pavers, joint cutters, joint sealers, pile drivers, truck cranes, and core drills) and personnel operations used in each construction stage. During the clearing and demolition phase, drones use high-definition cameras to confirm the safe distance between the demolition area's warning line and high-voltage lines, identify major hidden dangers such as illegal blasting and the dangerous placement of temporary facilities, and capture general hidden dangers such as inadequate pipeline protection and lack of warnings. During the roadbed construction phase, drones utilize thermal imaging and visible light fusion to identify the step-by-step construction of high slopes and soft soil treatment during the filling process; to detect the isolation of grader work areas and personnel intrusion during the paving process; to monitor the reversing command of road rollers and unauthorized entry into blind spots during the compaction process; to identify dangerous areas and illegal passage of machinery at inspection points during the inspection process; to inspect deep foundation pit support and perimeter protection during the drainage ditch process; and to verify the foundation depth of retaining walls and the parking of heavy machinery during the slope protection process. During the road base construction phase, drones use lidar to detect the risk of collisions between pavers and transport vehicles and traffic control during the paving process, monitor the reversing command of road rollers and unauthorized personnel entry during the compaction process, and identify illegal opening of traffic and geotextile coverage during the maintenance process. During the road surface construction phase, the drone's thermal imager captures in real time whether there are high-temperature burns in the asphalt paving and whether the mixture segregates. At the same time, the high-definition camera identifies blind spots of the roller and unauthorized personnel in the compaction process. It monitors the risk of mechanical damage to the cutting machine and high-temperature burns from the sealant in the joint cutting and sealing process. It also checks the inspection area for warning settings and potential injuries from the core drilling machine in the inspection process. During the construction phase of traffic safety and electromechanical systems, drones use zoom cameras to inspect for risks of pile driver overturning, ensure proper hoisting command, confirm the secure transport and binding of guardrail panels, ensure the compaction of foundation concrete, and verify whether the depth of the posts meets standards. Throughout the inspection process, the drones adaptively optimize their hovering time based on the regulatory needs of different construction stages, the severity and number of potential hazards, and transmit the collected hazard data back to the backend for analysis and calculation in real time.

[0046] Table 1. Items requiring supervision of drones during the site clearing and demolition phase.

[0047] Table 2. Items requiring supervision of drones during the roadbed construction phase.

[0048] Table 3. Items requiring supervision of drones during the road base construction phase.

[0049] Table 4. Contents requiring supervision of drones during the road surface layer construction phase.

[0050] Table 5. Items Requiring Supervision of Drones During the Traffic Safety and Electromechanical Construction Phase

[0051] 2. Construct a dynamic method for dividing inspection sub-zones based on the supervision needs of each construction stage. In highway construction scenarios, after drones take off from their nests, they fly sequentially to each construction section and hover for inspection. However, due to the limitation of the drone's maximum endurance, a single inspection cannot cover the entire construction section. Therefore, it is necessary to divide the entire road section into inspection sub-areas based on the maximum endurance.

[0052] To minimize the number of drones required and reduce equipment investment and operating costs while meeting the regulatory needs of each construction phase, the coverage area of ​​each sub-zone should be expanded as much as possible within the limits of flight endurance, allowing a single flight to inspect more sections. It is important to note that the regulatory requirements for drones differ across construction phases, resulting in variations in hovering time. Therefore, when dividing inspection sub-zones, it is essential to first obtain the current construction progress information for each section and, based on the preset hovering time range for each construction phase (as shown in Table 6), calculate the cumulative maximum hovering time for each section. This ensures that even under the most stringent task requirements (i.e., all sections operate at their respective maximum hovering times), the total cumulative time does not exceed the drone's maximum flight endurance, thus guaranteeing the feasibility and safety of sub-zone division.

[0053] The specific steps for division are as follows: Starting from the location of the drone nest (located near temporary facilities), the cumulative calculation is performed segment by segment along the construction route. The cumulative items include: the maximum hovering time of each segment, the flight time between adjacent segments (calculated from the drone inspection speed and the segment spacing), the flight time from the drone nest to the first segment, and the flight time from the last segment back to the drone nest. When the cumulative total time first approaches but does not exceed the maximum endurance of the drone, the boundary of the current sub-region is defined, as shown in equation (1). The remaining undefined segments continue to be divided, and the above process is repeated until the entire road segment is divided.

[0054] (1); in, For the flight time of the drone to the first segment; The number of regulatory sections within a sub-region; For the first Maximum hovering time for each segment; For the first Section 1 to 2 Flight time for each segment; The flight time for the drone to return to base; The maximum flight time of the main inspection drone.

[0055] This method can dynamically adjust the sub-area range according to the construction progress, maximizing the use of endurance while meeting the supervision needs of different construction stages. This reduces the number of drones required, lowers operating costs, and effectively avoids drones returning midway or mission interruption due to excessive flight distance, or wasting endurance due to insufficient flight distance.

[0056] To achieve accurate identification of potential hazards, drone inspections are typically set at a height of 40-50 meters. At this height, drones can cover an area of ​​50 meters (cross-sectional width) × 70 meters (road section length). Therefore, this area is used as a segment, and segments are continuously deployed along the construction section to achieve full coverage monitoring.

[0057] During the dynamic division of sub-zones, the monitoring sections continuously deployed along the construction route are divided into 70-meter units. For road sections that span construction stages and are less than 70 meters in length, the monitoring requirements and hovering times differ between different construction stages. Therefore, such road sections are monitored as independent sections and are not merged with other road sections.

[0058] Table 6 Recommended values ​​for hovering time range of single-segment UAVs

[0059] Case 1 (Case Study on Inspection Sub-zone Division Based on Supervision Needs at Each Construction Stage): The maximum flight time of drones is usually in the range of 45-59 minutes. A drone with a maximum flight time of 45 minutes is selected. As shown in Table 7, the construction progress of a certain 5-kilometer highway is as follows: 0-1km roadbed construction, 1-2km pavement base construction, 2-3km pavement surface layer construction, 3-4km pavement surface layer construction, and 4-5km traffic safety and electromechanical construction. Based on Table 8, the available hovering time for the UAV (maximum endurance minus round-trip and route flight time) is calculated to be 1980 seconds. As shown in Table 9, by accumulating the maximum hovering time segment by segment, it was found that when the accumulated time reached the 27th segment (water-stabilized construction), the accumulated time was exactly 1980 seconds. At this time, the sub-segment length was 1840 meters, that is, the first sub-segment covered 0-1840 meters (including 0-1km of roadbed and part of 1-1.84km of water-stabilized road). The remaining segments were divided into the second sub-segment. As the construction progresses over time, the hovering time of the same road segment will change at different times. Therefore, the sub-segment division needs to be recalculated regularly (e.g., once a week) to adapt to changes in the construction progress.

[0060] Table 7. Case studies of construction stages for different road sections

[0061] Table 8. Case Study of Usable Hovering Time Calculation for Drones

[0062] Table 9. Case Study of Cumulative Hovering Time Calculation in Different Sections

[0063] like Figure 8 As shown, based on the dynamic division method of inspection sub-zones, a certain road segment is divided into two sub-zones. Each sub-zone is equipped with one inspection drone (main drone), which can perform routine inspection tasks and, in the event of an emergency, interrupt routine tasks to prioritize flying to the incident location for emergency inspection. Simultaneously, each sub-zone is additionally equipped with a backup drone to quickly take over when the main drone's battery is insufficient to complete the remaining tasks or in the event of an emergency, ensuring uninterrupted monitoring of the entire road segment. During the specific flight process, the main drone first flies from its nest to the first segment of its sub-zone, hovers for a preset time, then proceeds to subsequent segments and hovers there, finally returning to the nest from the last segment.

[0064] 3. An Adaptive Optimization Method for Drone Hovering Time Based on the Monitoring Needs of Hidden Danger Incidents in Sections After the drone inspection sub-areas are divided, the hovering time needs to be dynamically optimized based on changes in potential hazards in each section. During the inspection, if the number of potential hazards increases in a certain section, the hovering time for that section should be appropriately increased (not exceeding the maximum hovering time) to achieve detailed identification of hazards and ensure that critical issues are not overlooked; if the number of potential hazards decreases, the hovering time should be reduced accordingly (not less than the minimum hovering time).

[0065] Therefore, an adaptive optimization method for drone hovering time based on the monitoring needs of potential hazards in a section is proposed. The hovering time of each inspection is iteratively updated based on the situation of the previous inspection, thereby achieving adaptive optimization as the risk status of the construction site changes in real time.

[0066] The core of this method is iterative optimization. During the first inspection by the drone, the hovering time of each section is uniformly set to the minimum hovering time of the corresponding construction stage to meet the minimum regulatory requirements. In subsequent iterative optimization processes, the hovering time is continuously optimized based on the previous hovering time. When the number of potential hazards increases, the hovering time is iteratively optimized in order of priority from high to low event weight, and all optimization results do not exceed the maximum hovering time preset for the construction stage.

[0067] (1) Iteration of hovering time when the number of potential incidents increases Let the first Before the next inspection, the hovering time of a certain section was: This time, a major new safety hazard has been added. Individual, general safety hazards One, non-standard operation One; iterate through each potential hazard event in order of priority, calculating the increment for each iteration. As shown in equation (2): (2); in, Weighting of currently handled potential safety hazards (recommendations for major safety hazards) General safety hazard recommendations Recommendations for non-standard operations ); Maximum hovering time; The hover time before iteration.

[0068] The current hover time is updated immediately after each iteration. After all new potential events are processed in sequence, the final hover time is calculated as shown in equation (3): (3); in, For the first The hovering time during the next inspection; For the first The hovering time during the next inspection; They represent the first One major safety hazard, the first A general safety hazard, the first The specific incremental value generated during the iteration process by an irregular operation. Record this incremental value for use in subsequent reductions.

[0069] (2) Deduction of hovering time when the number of potential incidents decreases Let the first During the second inspection, it was confirmed that major safety hazards had been reduced. Individual, general safety hazards One, non-standard operation One. From the set of incremental values ​​of similar hidden danger events in the historical records of this section, select the smallest one. Each increment value is summed and then subtracted from the current hover time, as shown in equation (4): (4); in, The smallest in history The sum of the incremental values; The smallest in history The sum of the incremental values; The smallest in history The sum of the incremental values.

[0070] To ensure a smooth decrease in hover time as the number of potential incidents decreases, and to avoid drastic jumps in hover time due to a large one-time deduction of a large increment, a deduction strategy of "prioritizing the deduction of the smallest increment" is adopted. This is because, as the hover time increases, the increment value generated in each subsequent iteration will naturally decrease as the current hover time gradually approaches the maximum hover time (i.e., the increment corresponding to later potential incidents is smaller). At this time, prioritizing the deduction of the smaller increment in the historical records can make the decrease in hover time more gradual, thereby achieving a smooth transition and ensuring the stability and controllability of the hover time adjustment process.

[0071] Case 2 (Optimization of Drone Hovering Time Based on the Need for Monitoring Hidden Dangers in a Section): Taking the hovering time of a certain section of the road surface layer as an example, =60 seconds, =150 seconds, =60 seconds. Assuming the first day's inspection (drone inspection once a day) reveals 3 new major safety hazards, 3 general safety hazards, and 3 non-standard operations in this section, the hovering time for the second day's inspection should iteratively increase based on the first day's hovering time, according to the priority of the hazard events. For example... Figure 13As shown, the hovering time of the section during the second day's inspection can be calculated, as well as the incremental value contributed by each major safety hazard, general safety hazard, and non-standard operation. If the second day's inspection finds that the section has reduced by 2 major safety hazards, 2 general safety hazards, and 2 non-standard operations, then the hovering time of the third day's inspection should be deducted from the second day's hovering time: select the two smallest increments caused by major safety hazards, the two smallest increments caused by general safety hazards, and the two smallest increments caused by non-standard operations from the historical records, sum these selected increments, and subtract them from the second day's hovering time to obtain the third day's hovering time, as shown in equation (5).

[0072] (5); Table 10 Case Study on Hovering Time Calculation of Inspection Drones on the Second Day

[0073] 4. Construct a dynamic linkage scheduling algorithm for UAVs based on the occurrence and termination of emergencies. 4.1 Emergency Dispatch of Unmanned Aerial Vehicles Based on Sudden Events like Figure 9 As shown, when emergencies such as slope landslides, personnel injuries, or construction vehicles illegally entering the work area occur during highway construction, the drone dynamic linkage scheduling algorithm is used for emergency dispatching of drones to achieve real-time detection of emergencies; specifically, it is divided into the following three situations: Scenario 1: The inspection drone (main drone) is in the nest. When an emergency occurs during highway construction, the first step is to determine that the main drone in the sub-area of ​​the incident is in standby mode. At this point, the main drone is directly dispatched to the scene of the emergency. Upon arrival, the main drone immediately and continuously monitors the emergency, determining in real time every second whether the emergency has ended. If the emergency has ended, the main drone immediately performs a return-to-home operation, and the entire process ends. If the emergency has not ended, the drone further determines every second whether its remaining battery power is sufficient for a return-to-home operation. If the battery power is only sufficient for a return-to-home operation, the drone immediately returns to home (the emergency is no longer monitored). If the battery power is sufficient, the drone continues to monitor the scene until the emergency has ended before returning to home, and the process finally terminates.

[0074] Scenario 2: The main drone is not in its nest and has insufficient power; a backup drone is dispatched for emergency response. When an emergency occurs during highway construction, if the main drone in the sub-area of ​​the incident is not in its nest (performing a routine inspection task) and its remaining battery power is insufficient to meet the power threshold requirement for the emergency (duration of flight to the emergency), then the incident will occur. Minimum detection time (Recommended 5 minutes) and return flight time (The sum of the two); At this time, the main drone continues to perform its original routine inspection tasks, while the backup drone is dispatched to take off and fly to the scene of the emergency. After the backup drone arrives at the scene, it continues to carry out detection work, and determines every second whether the event has ended: If the event has ended, the backup drone performs a return operation and the process ends; If the event has not ended, it determines every second whether the backup drone has enough remaining power to return. If the power is insufficient, it returns directly (the emergency will no longer be detected), and if the power is sufficient, it continues to maintain on-site detection until the event ends and it returns, and the process ends.

[0075] Scenario 3: The main drone is not in its nest but has sufficient power, so it interrupts its regular inspection mission for emergency dispatch. When an emergency occurs during highway construction, the system determines that the main drone in the sub-region where the incident occurred is not in its nest (it is performing a routine inspection task), and that its remaining battery power is sufficient to meet the emergency event's power threshold requirement (duration of flight to the emergency event). Minimum detection time (Recommended 5 minutes) and return flight time (The sum of the two); At this time, the main UAV interrupts its original routine inspection mission and flies directly to the scene of the emergency. After arriving, the main UAV continuously monitors the emergency, determining every second whether the event has ended: If the event has not ended, it determines every second whether the main UAV's remaining battery power is just enough to return. If the battery power is insufficient, it returns directly (the emergency is no longer monitored), and at the same time, a backup UAV is dispatched to take over the main UAV to complete the remaining routine inspection mission. If the battery power is sufficient, it continues to monitor the scene; If the event has ended, it further determines whether the main UAV's remaining battery power is sufficient to meet the needs of inspecting the remaining section and returning to the nest. If the battery power is insufficient, the main UAV returns and a backup UAV is dispatched to take over the remaining inspection mission. If the battery power is sufficient, the main UAV continues to complete the routine inspection mission of the remaining section, and the process ends.

[0076] 4.2 Unmanned Aerial Vehicle Inspection Path Planning Based on the End of an Emergency like Figure 10 As shown, after the emergency ends, if the remaining power of the main UAV is sufficient to meet the inspection and return requirements of the remaining section, the UAV dynamic linkage scheduling algorithm is used to plan the path for the interrupted regular inspection task, so as to complete the inspection task of the remaining section in the shortest possible time. The algorithm divides the interruption scenario into three categories: current section interruption, inspected section interruption, and uninspected section interruption, based on the relationship between the inspection position of the UAV at the time of the emergency and the current inspection progress.

[0077] For the current segment interruption scenario, such as Figure 11 As shown, the drone continues to advance from the current section and completes the inspection tasks of the remaining uninspected sections in sequence; For scenarios where inspected sections are interrupted, such as Figure 12As shown, the drone is dispatched to move forward and complete the inspection tasks of the uninspected sections in sequence at the interruption points; For scenarios involving interruptions in uninspected sections, such as Figure 13 As shown, the total time for the "forward" and "backward" paths is calculated separately, with the total time for the forward path being... To get from the emergency section at speed v Time to fly to the last segment The time it takes to fly back from the last segment to the interrupted segment. Time to fly back to the nest from the interrupted section The sum of the total time taken for the backward turnaround path. To get from the emergency section at speed v Time to fly to the interrupted section The time it takes to fly back from the interrupted section to the last section. Time to fly back to the nest from the last segment By comparing the total time taken by the two paths, the optimal inspection path with the shorter time is selected, enabling the drone to complete the inspection of the remaining sections in the shortest time, and ultimately ensuring the completion of the inspection task for the entire road section.

[0078] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. An adaptive planning method for unmanned aerial vehicle (UAV) inspection for highway construction supervision, characterized in that, Includes the following steps: S1. Propose drone inspection and supervision content for highway construction. Taking the construction scenario of new highway subgrade and pavement as the application object, the construction process is divided into five construction stages: clearing and demolition, subgrade construction, pavement base course construction, pavement surface course construction, and traffic safety and electromechanical construction. The drones perform inspection tasks to monitor the key equipment and personnel operation behavior in each construction stage in real time. According to the severity of the hidden dangers, they are divided into three categories: major safety hazards, general safety hazards, and non-standard operations. The sections along the highway construction route are divided into sections based on the coverage range of a single drone hovering as the basic detection unit. S2. Construct a dynamic division of inspection sub-zones based on the supervision needs of each construction stage. With the maximum flight time of the drone as a constraint, starting from the location of the drone nest, the hovering time of each segment, the flight time between adjacent segments, the flight time from the drone nest to the first segment, and the flight time from the last segment back to the drone nest are accumulated segment by segment along the construction route. When the total accumulated time first approaches but does not exceed the maximum flight time of the drone, the boundary of the current sub-region is defined. The remaining undefined segments are divided successively until the entire road segment is covered. S3. Adaptive optimization of drone hovering time based on the needs of monitoring potential hazards in specific areas. When the drone is first inspected, the hovering time of each section is set to the minimum hovering time of the corresponding construction stage. In subsequent iterations and optimizations, the hovering time is continuously optimized based on the previous hovering time. When the number of potential safety incidents increases in a certain section, the system iterates sequentially according to the priority order of major safety hazards, general safety hazards, and non-standard operations, calculating the hovering time increment and updating it each time. When the number of potential hazards in a certain section decreases, the smallest values ​​are selected from the incremental contribution values ​​of similar events in the historical records of that section and deducted. S4. Dynamic scheduling of drones based on the occurrence and termination of emergencies. S41. When an emergency occurs during highway construction, an emergency dispatch strategy shall be adopted based on the current status and remaining power of the main UAV. S42. After the emergency ends, the interruption scenarios will be classified according to the relationship between the current position of the drone and the inspection progress, and the optimal inspection path will be planned for each scenario.

2. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, Each designated sub-area is equipped with one main inspection drone and an additional backup drone to quickly take over when the main drone's power is insufficient to complete the remaining tasks or in the event of an emergency.

3. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, The monitoring equipment carried by the drone includes a high-definition zoom camera, a thermal imager, and a lidar.

4. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, The hovering time of each section calculated in step S2 is the maximum hovering time corresponding to each section, which is selected based on the preset hovering time range of each construction stage.

5. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, The formula for dynamically dividing the inspection sub-area in step S2 is as follows: ； in, The flight time for the drone to reach the first segment; The number of regulatory sections within a sub-region; For the first Maximum hovering time for each segment; For the first Section 1 to 2 Flight time for each segment; The flight time for the drone to return to base; The maximum flight time of the main inspection drone.

6. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, In step S3, during the initial inspection of the drone, the hovering time for each section is set to the minimum hovering time for the corresponding construction stage, based on the preset hovering time range for each construction stage.

7. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, The hovering time iteration rule when a potential hazard event increases in step S3 is as follows: Let the first Before the next inspection, the hovering time of a certain section was: This time, a major new safety hazard has been added. Individual, general safety hazards One, non-standard operation One; iterate through each potential hazard event in order of priority, calculating the increment for each iteration, as shown in the following formula: ； in, The weight of the potential hazards currently being addressed; Maximum hovering time; The hovering time before the iteration; The current hover time is updated immediately after each iteration. After processing all newly added potential events, the final hover time is calculated using the following formula: ； in, For the first The hovering time during the next inspection; For the first The hovering time during the next inspection; They represent the first One major safety hazard, the first A general safety hazard, the first The specific incremental value generated during the iteration process by an irregular operation.

8. The adaptive planning method for UAV inspection in highway construction supervision according to claim 7, characterized in that, The hovering time deduction rules for reducing potential hazards in step S3 are as follows: Let the first During the second inspection, it was confirmed that major safety hazards had been reduced. Individual, general safety hazards One, non-standard operation One; from the set of incremental values ​​of similar hidden danger events in the historical records of this section, select the smallest one respectively. The increments are summed and then subtracted from the current hover time, using the following formula: ； in, The smallest in history The sum of the incremental values; The smallest in history The sum of the incremental values; The smallest in history The sum of the incremental values.

9. The adaptive planning method for UAV inspection in highway construction supervision according to claim 2, characterized in that, Step S41, emergency dispatch based on sudden events, includes three dispatch scenarios: Scenario 1: When an emergency occurs, the main drone is in the nest. Directly dispatch the main drone to the scene of the emergency. Real-time judgment is made on whether the event has ended. If the event has ended, the return operation is performed. If the event has not ended, it is judged whether the remaining battery of the main drone is just enough to return. If the battery is just enough to return, it returns immediately. If the battery is sufficient, it continues to maintain the on-site monitoring status until the event ends before returning. Scenario 2: When an emergency occurs, the main drone is not in its nest and its remaining battery power is insufficient to meet the emergency's power threshold requirements. The main drone continues to perform its regular inspection tasks, while a backup drone is dispatched to take off and fly to the emergency site. The backup drone will determine in real time whether the emergency has ended. If the emergency has ended, the backup drone will return to base. If the emergency has not ended, the backup drone will determine whether its remaining battery power is sufficient for a return. If the battery power is insufficient, the backup drone will return directly. If the battery power is sufficient, the backup drone will continue to monitor the site until the emergency ends and then return to base. Scenario 3: When an emergency occurs, if the main drone is not in its nest and its remaining battery power is sufficient to meet the emergency's power threshold requirements, the routine inspection mission is interrupted, and the drone prioritizes flying to the emergency site for testing. If the emergency is not over, determine if the main drone's remaining battery power is only enough for a return trip. If the battery power is insufficient, the main drone returns directly, and a backup drone is dispatched to take over the remaining routine inspection mission. If the battery power is sufficient, on-site testing continues. If the emergency has ended, determine if the main drone's remaining battery power is sufficient to meet the requirements for inspecting the remaining section and returning to the nest. If the battery power is insufficient, the main drone returns, and a backup drone is dispatched to take over the remaining inspection mission. If the battery power is sufficient, the main drone continues to complete the remaining routine inspection mission.

10. The adaptive planning method for UAV inspection in highway construction supervision according to claim 1, characterized in that, Step S42, the inspection path planning after the emergency ends, is divided into three optimization strategies based on the drone interruption scenario: Current section interruption scenario: Dispatch the drone to continue moving forward from the current section and complete the inspection tasks of the remaining uninspected sections in sequence; Scenario of interruption in inspected section: Dispatch the drone forward and complete the inspection tasks of the uninspected section in sequence at the interruption point; In the scenario of interruption of uninspected sections: calculate the total time for both the "forward" and "backward" paths, and select the optimal path with the shortest time to complete the remaining inspection tasks; The total time for the "forward" path is the sum of the flight time from the emergency segment to the final segment, the flight time from the final segment to the interrupted segment, and the flight time from the interrupted segment to the nest; the total time for the "backward" path is the sum of the flight time from the emergency segment to the interrupted segment, the flight time from the interrupted segment to the final segment, and the flight time from the final segment to the nest.

Images